Rob Sherwood, and SDR transmitter performance

[Terry Fox]

BTW, I should point out that Rob Sherwood is very well respected, and many 
people follow his testing with baited-breath.  He also operates the radios 
that he tests during contests and at other times, giving him valuable 
real-world experience, which sometimes amplifies on, or changes his results.

I believe that he is among many people that think the "standard" methods for 
testing receivers will need changing, as SDRs bring a whole new set of 
challenges, problems, and benefits.  Quantifying actual receiver performance 
may differ drastically between "traditional" receivers and SDRs, and new 
testing/quantifying methodologies need to be created.

I also forgot to follow-up on the mention of "Pure Signal" on the SDR 
transmitter side.  Some hams are concerned about poor transmitter 
performance of certain SDR transmitters.  This may be justified, depending 
on the hardware and user setup.  Using the older SDR technologies, such as a 
QSD/QSE-based rig, there is a significant possibility of unwanted signals 
being created (or not adequately suppressed) and transmitted.  The 
most-often issue is poor image rejection, similar to a poorly-calibrated 
analog phasing transmitter.  Another is LO carrier leakage.  Neither of 
these can be removed with bandpass filters, as they are typically very close 
to the wanted transmitted signal.  Good balance of the I/Q channels, and 
sufficient carrier suppression is not always easy, especially without decent 
test equipment.  A spectrum analyzer is REAL handy here.  Audio noise (or 
Windows sounds!) from a poor quality sound card can also generate 
significant wide-band garbage, along with unwanted spikes at 
sampling-related frequencies.  Again, this kind of crud is extremely 
difficult to remove by filtering.

The problems above are much less of an issue with Digital Up 
Converter(DUC)-based SDR transmitters.  The phasing and adjusting are all 
done while the transmitted signal is still in the digital (number) realm, 
and as long as careful planning has been done with the DUC such that the 
numbers representing the signal do not fall into certain 
number-representation pitfalls, the final "signal" sent to the D/A converter 
"should" be pretty clean.

The HPSDR folks (one in particular) have been working on a method of 
pre-distorting the transmitted signal while it is still represented as 
numbers inside the DUC/SDR blocks, to effectively counteract distortions 
that would otherwise occur to the signal.  When I was involved in the DTV 
transition, Harris did this with digital television transmitters, and they 
called it "Digital Adaptive PreDistortion".  I'm not sure if another DTV 
transmitter manufacturer already had that term under lock or not, but the 
result is the same.  If you have a receiver tuned to your transmitted 
signal, and monitor that signal for distortions coming up, you can insert a 
"correction" signal to offset the distortion.  Of course, it's much harder 
than that in practice.

These problems are not directly associated with any of the usual transmitter 
distortions and unwanted signals that occur during amplification of the 
transmitted RF signal.  You still need the typical low-pass filters, etc 
these days once an analog amplifier chain is included.  It would be 
interesting to come up with an SDR transmitter that directly creates a 
powerful RF signal right out of a high-power D/A converter!  But then, you 
would still need analog filters to remove clocking artifacts (at a minimum). 
A 100W D/A converter would be something!

Anyway, that's more ramblings from Charleston.
73, Terry, WB4JFI